ARTICLE

HGF-induced migration depends on the PI(3,4,5)P3- binding microexon-spliced variant of the Arf6 exchange factor cytohesin-1

Colin D.H. Ratcliffe1,2, Nadeem Siddiqui1,2, Paula P. Coelho1,2, Nancy Laterreur1, Tumini N. Cookey1, Nahum Sonenberg1,2, and Morag Park1,2,3,4

Differential inclusion or skipping of microexons is an increasingly recognized class of alternative splicing events. However, the functional significance of microexons and their contribution to signaling diversity is poorly understood. The Met Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 receptor tyrosine kinase (RTK) modulates invasive growth and migration in development and cancer. Here, we show that microexon switching in the Arf6 guanine nucleotide exchange factor cytohesin-1 controls Met-dependent cell migration. Cytohesin-1 isoforms, differing by the inclusion of an evolutionarily conserved three-nucleotide microexon in the pleckstrin homology domain, display differential affinity for PI(4,5)P2 (triglycine) and PI(3,4,5)P3 (diglycine). We show that selective phosphoinositide recognition by cytohesin-1 isoforms promotes distinct subcellular localizations, whereby the triglycine isoform localizes to the plasma membrane and the diglycine to the leading edge. These data highlight microexon skipping as a mechanism to spatially restrict signaling and provide a mechanistic link between RTK-initiated phosphoinositide microdomains and Arf6 during signal transduction and cancer cell migration.

Introduction The Met receptor tyrosine kinase (RTK) coordinates inva- nositol 4,5-bisphosphate (PI(4,5)P2) in the cell, however this lipid sive growth in response to its ligand hepatocyte growth factor may be modified by phosphoinositide 3-kinase (PI3K) to locally

(HGF). This is tightly regulated during development to promote generate phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P3; a morphogenic program that is essential for liver development Whitman et al., 1988), thereby affecting local recruitment and and migration of muscle precursors into the limb bud as well activity of possessing PI(3,4,5)P3-binding domains. as wound healing in the adult (Gherardi et al., 2012). Under The small GTPase Arf6 is critical for Met-dependent invasive pathophysiological conditions, deregulated signaling by the Met growth, although the molecular mechanisms that link Met to RTK leads to enhanced cell migration and metastatic spread of Arf6 activation in cancer cells are unknown (Tushir and D’Souza- cancer cells (Gherardi et al., 2012; Parachoniak and Park, 2012; Schorey, 2007). Arf small GTPases are members of a superfamily Knight et al., 2013). of molecular switches that mediate changes in cell morphol- The proinvasive properties of Met are tightly regulated by ogy, endomembrane traffic, and cell signaling (Gillingham and spatial localization of signaling complexes on subcellular com- Munro, 2007; Simanshu et al., 2017). Arf6 is unique among Arf partments, including dorsal ruffles, invadopodia, lamellipodia, proteins in that it localizes primarily to the plasma membrane and endosomes (Maroun et al., 1999a,b; Palamidessi et al., 2008; and endosomes, as opposed to Arf1 and Arf3, which localize pre- Abella et al., 2010; Rajadurai et al., 2012; Ménard et al., 2014). dominantly to the Golgi apparatus (Donaldson and Jackson, 2011). Each of these compartments possesses distinct morphological Active, GTP-bound Arf6 modulates processes that are critical for and molecular features. While Met recruits many different effec- cell migration and tumor metastasis (Muralidharan-Chari et al., tors, not all complexes are assembled at each subcellular location 2009; Yoo et al., 2016). These include endosomal recycling of the where Met is active and additional determinants must define the Met RTK or integrin receptors, clathrin-independent endocyto- localization of different signaling complexes. For example, the sis, and rearrangement of the actin cytoskeleton (Powelka et al., plasma membrane is the predominant source of phosphatidyli- 2004; Eyster et al., 2009; Parachoniak et al., 2011; Ratcliffe et al.,

1Rosalind and Morris Goodman Cancer Research Centre, McGill University, Montreal, Quebec, Canada; 2Department of Biochemistry, McGill University, Montreal, Quebec, Canada ; 3Department of Medicine, McGill University, Montreal, Quebec, Canada ; 4Department of Oncology, McGill University, Montreal, Quebec, Canada.

Correspondence to Morag Park: morag.park@mcgill​ .ca​ .

© 2018 Ratcliffe et al. This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://​www​.rupress​.org/​terms/​). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://​creativecommons​.org/​licenses/​by​-nc​-sa/​4​.0/​).

Rockefeller University Press https://doi.org/10.1083/jcb.201804106 285 J. Cell Biol. 2018 Vol. 218 No. 1 285–298 2016). Like the majority of small GTPases, Arf6 cycles between an “off” GDP-bound state and an “on” GTP-bound state. Cycling be- tween these states is enhanced by guanine nucleotide exchange factors (GEFs), which promote GDP release, and GTPase activa- tion proteins, which promote hydrolysis of GTP (Donaldson and Jackson, 2011; Simanshu et al., 2017). Subcellular localization of GEF and GTPase activation proteins imposes spatiotemporal reg- ulation on small GTPase signaling, restricting activity to specific subcellular locations. However, the in vivo subcellular determi- nants for Arf6 activation remain to be fully defined. Arf GEFs fall into seven families, three of which encompass putative Arf6 GEFs, including cytohesin (1–4), IQSEC​ (1–3), and PSD (1–4; Casanova, 2007; Donaldson and Jackson, 2011). These families are defined by the presence of a Sec7 domain that en- hances the release of GDP from Arf proteins. In addition, there

are multiple splice variants of Arf GEFs. Some differ by an entire Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 domain, whereas others involve microexons (Ogasawara et al., 2000; Fukaya et al., 2016). The best-characterized microexon splice variants are two isoforms of cytohesin-2. These isoforms differ by a 3-nt microexon, whose splicing leads to an additional glycine residue within the cytohesin-2 pleckstrin homology (PH) domain (Cronin et al., 2004), yet a functional difference for these has not been determined. Microexons are a recently described class of exons that are ≤27-nt long and predominantly found in structured regions of proteins. Microexons are frequently identified in brain-derived transcripts, including cytohesin-2, and are alternatively reg- ulated in individuals with autism spectrum disorder (Irimia et al., 2014). However, their functional roles in other diseases, par- ticularly cancer, are unknown. Furthermore, despite decades of research on signaling from RTKs, such as Met, the role of alter- native splicing in generating multiple isoforms of downstream effectors is also largely unexplored. Here, we demonstrate differ- ential functions for cytohesin-1 isoforms whereby a splice variant of cytohesin-1 that lacks a 3-nt microexon has distinct phospho- lipid binding and is uniquely required for HGF-dependent cell migration. Thus, we provide a mechanistic understanding into the regulation of HGF-dependent cell migration by microexon skipping in a specific effector acting downstream of Met.

Results Cytohesin-1 regulates HGF-dependent cell migration Stimulation of epithelial and many cancer cells with HGF promotes activation of the Met RTK and subsequent cellular morphological changes leading to enhanced cell migration. Figure 1. Cytohesin-1 depletion reduces HGF-dependent cell migra- The migratory and invasive program induced by the Met RTK tion. (A–C) Random cell migration of HeLa cells (A), MDA-MB-231 cells (B), or HCC1143 cells (C) treated with the indicated siRNA smartpool without (−, specifically requires the small GTPase Arf6 (Tushir and D’Souza- unfilled) or with (+, filled) HGF.(D) Random cell migration of CYTH1 KO or Schorey, 2007). Using previously described siRNAs targeting Arf lentiCRI​SPR v2 empty vector HeLa clones treated with or without HGF. All family GTPases (Ratcliffe et al., 2016), we confirmed the role of quantified data indicate mean values ± SEM from three or four independent Arf6 in Met-dependent rearrangement of the actin cytoskeleton experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.00.1. (Fig. S1). However, the regulation of Arf6-dependent actin remodeling by Met is not fully understood. To identify Arf GEFs required for HGF-dependent cell migration, we measured Met-dependent migration and Arf6-dependent cell migration the expression of putative Arf GEFs, which are defined by the (Palamidessi et al., 2008; Parachoniak et al., 2011; Frittoli et al., presence of a Sec7 domain (Donaldson and Jackson, 2011). We 2014). Out of 10 putative Arf GEFs, six were detectably expressed used HeLa cells, which have been extensively studied for both in HeLa cells, as assessed by quantitative RT-PCR (Fig. S2 A).

Ratcliffe et al. Journal of Cell Biology 286 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 These include cytohesin-1, cytohesin-2, cytohesin-3, IQSEC1,​ Grp1], and cytohesin-4). These proteins consist of a coiled-coiled IQSEC2,​ and PSD3. domain, a Sec7 domain with Arf GEF activity, and a PH domain To assess the effect of Arf GEFs on cell migration, each Arf GEF that selectively recognizes phosphoinositides (Chardin et al., was independently reduced by siRNA-mediated silencing to ≤30% 1996). Two isoforms of cytohesin family members are expressed (Fig. S2 B), and cells were imaged every 15 min for 24 h in the pres- that differ by the inclusion of a evolutionarily conserved 3-nt ence and absence of HGF. Quantification of cell speed between 16 microexon resulting in an additional glycine residue in the PH and 24 h after stimulation revealed that silencing of cytohesin-1 domain (Ogasawara et al., 2000; Irimia et al., 2014). We refer to and IQS​EC2 reduced HGF-dependent cell migration to 36% and these isoforms as diglycine and triglycine variants (Fig. 2 A). The 46% of control cells, respectively (Fig. 1 A). In contrast, silencing microexons of cytohesin-1 and cytohesin-2 show similar inclu- of IQSEC1​ (also known as BRAG2) enhanced HGF-independent cell sion (or percent spliced in [PSI]) values in HeLa cells (76% and migration, with no further increase of cell speed observed follow- 69%) and HCC1143 cells (66% and 66%) but different PSI values in ing HGF treatment (Fig. 1 A). This enhancement in cell speed is MDA-MB-231 cells (93% and 45%; Fig. 2 B), whereas cytohesin-3 likely mediated by an increase in the cell surface levels of integrins shows a reduced PSI of 5% in HeLa and MDA-MB-231 cells and caused by IQSEC1​ silencing (Moravec et al., 2012). 21% in HCC1143 cells. To identify the splice variant of cytohesin-1 Whereas unstimulated IQS​EC2-depleted cells were more that mediates HGF-dependent cell migration, we generated a

spread and could readily be distinguished from control cells panel of stable cell lines expressing GFP-tagged isoforms or mu- Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 based on their morphology, cytohesin-1–depleted cells appeared tants of cytohesin-1 in the CYTH1 KO cell background (Fig. S4 morphologically indistinguishable from control cells but had A). Under these conditions the diglycine GFP-CYTH1, but not diminished HGF-induced cell migration. This implies that triglycine GFP-CYTH1, rescued HGF-dependent cell migration IQS​EC2 depletion may have a more general effect on the actin when compared with empty vector control (Fig. 2 C). Rescue was cytoskeleton, while the effects of cytohesin-1 silencing in these dependent on the nucleotide exchange activity of the diglycine cells are required for HGF/Met signaling and cell migration. isoform, as expression of a GFP-CYTH1 construct with a muta- Met and Arf6 activity have each been implicated in breast tion of a glutamic acid essential for exchange activity (E157K) cancer cell migration (Hashimoto et al., 2004; Knight et al., was unable to rescue HGF-dependent cell migration (Fig. 2 C). In 2013; Li et al., 2015). To establish if cytohesin-1 played a role addition to a decrease in cell migration in 2D, CYTH1 KO reduced in Met-dependent migration, the basal-like breast cancer cell HGF dependent invasion through a 3D collagen matrix, which lines MDA-MB-231 and HCC1143, which express cytohesin-1, could be rescued by expressing the diglycine, but not triglycine, cytohesin-2, and cytohesin-3, were investigated (Fig. S2 C). isoform of cytohesin-1 (Fig. 2, D and E). Rescue of HGF-induced Migration speed of both cell lines is increased following HGF cell invasion is similarly dependent on the exchange activity of stimulation (Fig. 1, B and C), consistent with previous observa- diglycine cytohesin-1, since the E157K cytohesin-1 mutant failed tions for MDA-MB-231 cells (Rajadurai et al., 2012). Silencing to rescue cell invasion in response to HGF (Fig. 2, D and E). Given cytohesin-2 in MDA-MB-231 cells increased cell migration, and these observations, we propose that the diglycine isoform of cy- there was a trend toward reduced HGF-dependent cell migration tohesin-1 acts downstream of the Met RTK and is required for in HCC1143 cells, supporting a role for cell-type–specific effects cancer cell HGF-dependent migration and invasion. for cytohesin-2. Silencing cytohesin-1 reduced HGF-dependent cell migration in both MDA-MB-231 and HCC1143 cell lines (Fig. 1, Cytohesin-1 splice variants differentially mediate B and C; and Fig. S2 D). Hence, we focused on cytohesin-1. The membrane ruffling decreased migratory phenotype observed by siRNA-mediated Cell migration requires the spatial coordination of multiple depletion of cytohesin-1 was validated by generating HeLa cells signals. Upon HGF stimulation, Met is rapidly internalized and with stable knockout (KO) of cytohesin-1 using the lentiCRISPR​ a fraction of these receptors are recycled to the leading edge, v2 system, with two independent guide RNAs targeting exon 2 where Rac1 is active and induces rearrangement of the actin cyto- of CYTH1 (Fig. S2 D). When HGF-dependent cell migration was skeleton (Royal et al., 2000; Palamidessi et al., 2008; Parachoniak compared in three control and CYTH1 KO clones, CYTH1 KO phe- et al., 2011; Ménard et al., 2014). Consistent with our data and nocopied siRNA-mediated silencing of cytohesin-1, whereby all previous reports, HGF induces a rearrangement in the actin cyto- clones displayed reduced HGF-dependent migration speed when skeleton with the formation of peripheral actin ruffles (Fig. 3 A). compared with control clones (Fig. 1 D). We did not observe a sig- Consistently, in CYTH1 KO cells, the percentage of cells with nificant difference in Met stability or Met, Akt, or Erk1/2phos- HGF-induced peripheral actin ruffles was reduced (51% in CTL phorylation between control and CYTH1 KO cells upon HGF versus 23% in KO). This could be rescued by expression of digly- treatment (Fig. S3). These data support a model in which cyto- cine GFP-CYTH1 (47%), but not the GEF exchange E157K mutant hesin-1 regulates HGF-dependent cell migration independently (22%; Fig. 3, A and C). Intriguingly, the majority of cells overex- of known downstream signaling targets and may represent a pressing the triglycine GFP-CYTH1 variant displayed peripheral novel pathway to Arf6 activation. actin ruffles (51%) in the absence of HGF stimulation that were not increased following HGF stimulation (Fig. 3, A and C). This Diglycine cytohesin-1 regulates HGF-dependent cell suggests that while the triglycine variant was unable to rescue migration and invasion HGF-dependent cell migration, overexpression of this isoform Cytohesin-1 belongs to a family of four proteins (cytohesin-1, is sufficient to promote downstream signals that enhance mem- cytohesin-2 [also known as ARNO], cytohesin-3 [also known as brane ruffling. Consistent with this, the GEF inactive E157K tri-

Ratcliffe et al. Journal of Cell Biology 287 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020

Figure 2. Diglycine cytohesin-1 regulates HGF-dependent cell migration and invasion. (A) Domain organization of CYTH1. (B) PSI values of cytohesin-1, cytohesin-2, or cytohesin-3 microexons in HeLa, MDA-MB-231, and HCC1143 cell lines. (C) Random cell migration of pLVX empty vector or eGFP-CYTH1 variant expressing cells treated with or without HGF. (D) Representative images of invasion of DAPI stained cell lines. (E) Quantification of experiments shown in D as distance from seeded area. Scale bar, 100 µm. All quantified data indicate mean values ± SEM from three independent experiments. *, P < 0.05; ***, P < 0.001. glycine mutant showed decreased ability to promote peripheral membrane protrusion velocity (82% of control) and maximum actin ruffles (16%; Fig. 3, A and C). These observations demon- displacement (104% of control). Rescue depended on the Arf GEF strate a significant difference in cytohesin-1 isoform function activity of diglycine cytohesin-1, since cells expressing diglycine and sensitivity to Met RTK stimulation. GFP-CYTH1 E157K failed to increase membrane velocity (35% of To quantitatively assess the effect of HGF on plasma mem- control) or maximum displacement (45% of control) relative to brane dynamics, we performed live-cell imaging in response to CYTH1 KO cells. HGF stimulation and assessed the relative position of the plasma While there was no significant effect of overexpressing the membrane every 15 s between 15 min and 60 min after HGF stim- triglycine isoform on net membrane velocity, we observed an ulation. In control cells, lamellipodia are observed in response to HGF-independent increase in the maximum displacement (193% HGF, with the formation of a leading edge that extends forward of control), indicating that these cells were actively ruffling but with a velocity of 0.081 µm/min and maximum displacement of failed to produce a stable leading edge (Fig. 3, B, D, and E). This ef- 5.74 µm, compared with 0.006 µm/min and 2.43 µm in the ab- fect was also dependent on the Arf GEF activity of cytohesin-1, as sence of HGF (Fig. 3, B, D, and E). In contrast, in CYTH1 KO cells, the E157K triglycine mutant failed to increase the maximum dis- both the velocity and maximum displacement of membrane pro- placement (Fig. 3, B and E). Together, this demonstrates that the trusions in response to HGF were reduced to 21.4% and 58% of diglycine PH domain mediates HGF-dependent cell migration and control cells, respectively (Fig. 3, B, D, and E). Consistent with establishment of a leading edge in a migrating cell and suggests its ability to rescue cell migration, diglycine GFP-CYTH1 rescued that phosphoinositide recognition regulates these processes.

Ratcliffe et al. Journal of Cell Biology 288 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020

Figure 3. Cytohesin-1 regulates membrane ruffling and actin cytoskeleton rearrangement. (A) Confocal images of cells counterstained with phalloidin (F-actin) and DAPI treated (−, unfilled; +, filled). Arrowheads indicate peripheral membrane ruffles. Scale bar, 20 (B)µm. Kymographs were generated from linescans of the cells’ leading edge imaged between 15 and 60 min after HGF treatment.(C) Quantification of experiments shown in A.(D and E) Quantification of experiments shown in B. All quantified data indicate mean values ± SEM from three or four independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Cytohesin-1 variants differentially recognize has a 640-fold greater affinity for I(1,3,4,5)P4 over I(1,4,5)P3. The phosphoinositide headgroups triglycine variant of cytohesin-1 PH domain binds to I(1,3,4,5)P4 The binding affinities of the PH domain of cytohesin-2 and cy- with an affinity of 3.03 µM (∼100-fold lower than diglycine) tohesin-3 for different inositol phosphate headgroups have been and I(1,4,5)P3 with an affinity of 7.23 µM (∼3-fold higher than extensively characterized (Klarlund et al., 2000; Cronin et al., diglcyine; Fig. 4 A). These results, similar to those observed for 2004). These studies show that the diglycine variant of cyto- cytohesin-2 by Cronin et al. (2004), support that the diglycine hesin-2 has a significantly stronger affinity (14-fold) for I(1,3,4,5) variant of cytohesin-1 preferentially interacts with PI(3,4,5)P3 P4, which is the headgroup of PI(3,4,5)P3, relative to I(1,4,5)P3, the on membranes. To test this, we titrated PI(3,4,5)P3-containing headgroup of PI(4,5,)P2 (Cronin et al., 2004), whereas the trigly- liposomes with the diglycine cytohesin-1 PH domain containing cine variant is less selective, binding to both with similar affini- the C-terminal polybasic region and found that it bound with ties. To characterize the specificity of cytohesin-1 for I(1,3,4,5)P4 a comparable affinity (Kd = 0.054 µM) to the headgroup alone and I(1,4,5)P3, we performed isothermal titration calorimetry (Fig. 4 C), confirming the ability of the diglycine PH domain to (ITC) using recombinant cytohesin-1 PH domain variants (Fig. recognize PI(3,4,5)P3 in the context of a lipid membrane. S4 B). We determined that the diglycine cytohesin-1 PH domain To gain further insight into the binding properties of the digly- bound to I(1,3,4,5)P4 with a Kd of 0.033 µM and I(1,4,5)P3 with a cine variant of cytohesin-1, we generated a homology model of its Kd of 21.05 µM (Fig. 4, A and B). This indicates that cythohesin-1 PH domain based on the structure of the PH domain of cytohesin-3

Ratcliffe et al. Journal of Cell Biology 289 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 and HGF-dependent cell migration, we compared CYTH1 KO cells expressing WT diglycine GFP-CYTH1 or the R280C mutant using F-actin staining and live-cell imaging. In response to HGF, cells expressing the diglycine GFP-CYTH1 (R280C) had a reduced capacity to form peripheral actin ruffles compared with cells expressing a WT diglycine GFP-CYTH1 (Fig. 5, A and C). The velocity and maximum displacement of the leading edge were also reduced in cells expressing the R280C diglycine mutant compared with WT GFP-CYTH1 (23% and 39% of WT, respectively; Fig. 5, B, D, and E). Consistent with this, HGF-dependent cell migration was reduced in cells expressing the diglycine GFP- CYTH1 (R280C) compared with WT (47% of control; Fig. 5 F), supporting that phosphoinositide engagement by diglycine cytohesin-1 is required for HGF-dependent cell migration. To establish if phosphoinositide binding was required for the

constitutive membrane ruffling induced by overexpression of the Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 triglycine variant, CYTH1 KO cells expressing WT triglycine GFP- CYTH1 or R281C (equivalent to diglycine R280C) were examined. Cells expressing the R281C mutant showed significantly fewer peripheral actin ruffles than WT (Fig. 5, G and I). When comparing the membrane dynamics of these cells, cells expressing triglycine GFP-CYTH1 R281C demonstrated reduced maximum displacement compared with WT (33.5% of WT triglycine; Fig. 5, H, J, and K). Hence, phosphoinositide recognition is a required step for both diglycine cytohesin-1–dependent membrane ruffling in response to HGF and constitutive membrane ruffling promoted by triglycine cytohesin-1.

Selective membrane recruitment of cytohesin-1 splice variants

The abundance of PI(4,5)P2 at the plasma membrane is approx- imately two orders of magnitude higher than that of PI(3,4,5)P3 Figure 4. G272 defines the phosphoinositide binding selectivity of (Stephens et al., 1991; Malek et al., 2017). However, HGF stimu- cytohesin-1. (A) Kd values measured by ITC. NB, no binding. (B) ITC trace of lation activates PI3K, increasing PI(3,4,5)P3 and recruitment of I(1,3,4,5)P4 titrated into diglycine CYTH1 PH domain. (C) ITC trace of diglycine PI(3,4,5)P3 binding proteins to PI(3,4,5)P3 microdomains (Maroun CYTH1 PH domain (aa 243–397) titrated into PI(3,4,5)P3-containing liposomes. et al., 1999b). Hence, to test if the diglycine variant of cytohesin-1 (D) Molecular model of the diglycine CYTH1 PH domain bound to I(1,3,4,5)P4 is specifically recruited to the plasma membrane upon PI3K ac- or I(1,4,5)P3. Error estimates shown are from fitting the data using a single site-binding model. tivation in response to HGF, cells expressing GFP-CYTH1 splice variants were stimulated with HGF for the indicated times, and membrane-bound cytohesin-1 was imaged by partially permea- (Fig. 4 D). Since there is an ∼90% sequence identity between the bilizing cells allowing for cytosolic GFP-CYTH1 to dissipate (Fig. PH domains of cytohesin family members, we anticipate that the S4 A and Video 1). Diglycine GFP-CYTH1 localized to the plasma phosphoinositide-binding pocket would be conserved. Based on membrane within 3 min after HGF stimulation, and recruitment previous studies, we predicted that R280 forms contacts with the was maintained for up to 60 min (Fig. 6 A). Notably, diglycine GFP-

3′ phosphate of I(1,3,4,5)P4 or 4′ phosphate of I(1,4,5)P3, and this CYTH1 membrane localization was polarized toward the leading site is required for a detectable interaction. Consistent with this edge of the cell upon HGF stimulation. Importantly, mutation of model, an R280C mutation when introduced into diglycine cyto- the phosphoinositide-binding pocket (R280C) abrogated recruit- hesin-1 PH domain abrogated any interaction with I(1,3,4,5)P4 or ment of diglycine GFP-CYTH1 to the leading edge (Fig. 6 B). In I(1,4,5)P3 (Fig. 4 A). Together, these data indicate that the diglycine contrast, the triglycine GFP-CYTH1 variant is constitutively asso- variant of cytohesin-1 specifically recognizes PI(3,4,5)P3, whereas ciated with the plasma membrane and is observed throughout the the triglycine variant may bind both PI(4,5)P2 and PI(3,4,5)P3, cell perimeter (Fig. 6 B). Recruitment was not further enhanced albeit with ∼3-fold higher and ∼100-fold lower affinities, respec- by HGF treatment, supporting a distinct mechanism of membrane tively, compared with the diglycine variant. recruitment that is distinct but also dependent on phosphoinos- itide binding, since a mutation in the phosphoinositide-binding Phosphoinositide binding of CYTH1 regulates membrane pocket (R281C) abrogated recruitment of triglycine cytohesin-1 ruffling and cell migration to the plasma membrane (Fig. 6 C). To test directly whether phosphoinositide recognition by To test whether plasma membrane recruitment of cytohesin-1 cytohesin-1 is required for the formation of cell protrusions was dependent on PI3K activity, cells expressing the diglycine or

Ratcliffe et al. Journal of Cell Biology 290 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020

Figure 5. Cytohesin-1 phosphoinositide binding is required for membrane ruffling and HGF-dependent cell migration. (A) Confocal images of cells coun- terstained with phalloidin (F-actin) and DAPI, with or without HGF. (B) Kymographs were generated from linescans of the cells’ leading edge imaged between 15 and 60 min after HGF treatment.(C) Quantification of experiments shown in A.(D and E) Quantification of experiments shown in B. (F) Random cell migration of diglycine EGFP-CYTH1 or R280C-ovexpressing cells treated with or without HGF. (G) Confocal images of cells counterstained with phalloidin (F-actin) and DAPI and treated with or without HGF (H) Kymographs were generated from linescans of the cells’ leading edge imaged between 15 and 60 min after HGF treatment. (I) Quantification of experiments shown in G.(J and K) Quantification of experiments shown in H. Scale bars, 20 µm. Arrowheads indicate periph- eral membrane ruffles. All quantified data indicate mean values ± SEM from three independent experiments. **, P < 0.01; ***, P < 0.001; n.s., not significant.

triglycine GFP-CYTH1 were pretreated with pan-PI3K inhibitors PI(3,4,5)P3 generated downstream from Met and growth factor Wortmannin and LY294002 and localization of GFP-CYTH1 signaling, the triglycine variant is constitutively recruited to the assessed. Pretreatment with either inhibitor abrogated HGF- plasma membrane in a PI(4,5)P2-dependent manner. Therefore, dependent membrane recruitment of diglycine GFP-CYTH1, the phosphoinositide-binding specificity of microexon- whereas localization of the triglycine GFP-CYTH1 to the plasma containing splice variants of CYTH1 defines the context for membrane was not significantly altered (Fig. 6 D). By comparing membrane ruffling, cell migration, and invasion. cells overexpressing the WT or catalytically dead PI(3,4,5)P3 lipid phosphatase (PTEN), we found that increased PTEN activity attenuated diglycine GFP-CYTH1 membrane recruitment (Fig. Discussion S5). These data support a requirement for PIP3 in diglycine Initiation of cellular signaling through activation of RTKs is well GFP-CYTH1 localization. Consistent with the increased recognized as a key event in cellular mitogenic or morphogenic affinity of triglycine GFP-CYTH1 for PI(4,5)P2, pretreatment response to growth factors. However, the cooperating molecular of cells with ionomycin to reduce plasma membrane levels of determinants of signal localization for these processes are still

PI(4,5)P2 (Botelho et al., 2000) abolished peripheral localization poorly understood. Arf proteins have been implicated in multiple of triglycine GFP-CYTH1 (Fig. 6 E). Together, these data show biological processes that involve membrane ruffling, including that cytohesin-1 splice variants are differentially recruited to the cancer cell invasion (Morishige et al., 2008), Met-dependent plasma membrane in vivo. Whereas the diglycine variant binds migration (Parachoniak et al., 2011), and bacterial invasion

Ratcliffe et al. Journal of Cell Biology 291 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020

Figure 6. Cytohesin-1 localization is defined by G272. (A) HeLa cells stably expressing EGFP-tagged diglycine CYTH1 were either not treated (−, unfilled) or treated (+, filled) with HGF for the indicated time points, permeabilized with ice-cold 0.05% saponin in Pipes buffer, and imaged by confocal microscopy. (B) HeLa cells stably expressing EGFP-tagged variants of CYTH1 were either not treated (−, unfilled) or treated (+, filled) with HGF for 15 min, permeabilized with ice-cold 0.05% saponin in Pipes buffer, and imaged by confocal microscopy.(C and D) Cells were prepared as in A, except they were pretreated

Ratcliffe et al. Journal of Cell Biology 292 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 (Humphreys et al., 2016). Activation of Arf proteins is modu- migration, suggesting the potential importance of microexons in lated by a family of Arf GEFs, yet specificity for distinct Arf GEF processes involved in metastatic behavior, and emphasizes the functions is still poorly understood. In our screen of Arf GEFs importance of developing a functional understanding of the roles that regulate cell migration in response to the proto-oncogenic of microexons in health and disease. RTK Met, we observed among six Arf GEFs expressed in HeLa Previous studies have shown that isoforms of cytohesin fam- cells, a specific requirement for cytohesin-1. Cytohesin proteins ily members interact selectively with PI(4,5)P2 or PI(3,4,5)P3 regulate phagocytosis in Dictyostelium discodium (Müller et al., (Klarlund et al., 2000; Cronin et al., 2004). However, a functional 2013), adhesion of lymphoid cells (El azreq and Bourgoin, 2011), distinction between microexon splice variants has not been ad- kidney repair following acute injury (Reviriego-Mendoza and dressed. Here, we show by ITC and confocal microscopy that Santy, 2015), and processes associated with cell migration. microexon splicing is a novel phosphoinositide switch whereby Previous evidence suggests specific roles for cytohesins in the the shorter cytohesin-1 isoform lacking a single glycine residue contexts of RTK signaling and integrin trafficking. For example, (diglycine) binds PI(3,4,5)P3 in vivo and the longer isoform (tri- cytohesin-3, but not cytohesin-2, acts downstream of Met to glycine) binds PI(4,5)P2. Thus, the diglycine cytohesin-1 variant, promote integrin recycling and angiogenesis in endothelial but not triglycine cytohesin-1, is a molecular link between Met cells (Hongu et al., 2015), whereas silencing cytohesin-2 or activation, PI3K signaling, Arf6, and HGF-dependent biology.

cytohesin-3 differentially affects β1 integrin trafficking in HeLa Initial data from brain tissue indicated that triglycine PI(4,5)P2- Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 cells (Oh and Santy, 2010). The localization of cytohesin-1 at the binding cytohesin-1 was the predominant isoform (Ogasawara periphery of the cell and known roles for Arf6 in endocytosis et al., 2000). However, recent analysis of RNA-sequencing data suggest that cytohesin-1 isoforms may differentially regulate from a variety of tissues indicates that PSI values for the cy- Met endocytosis and trafficking (Mayor et al., 2014). While tohesin-1 microexon can vary from 25% in liver and epithelial the different functions attributed to cytohesin proteins may be cells (predominantly diglycine) to >95% in muscle and white due to their differential expression (Ogasawara et al., 2000; Oh blood cells (predominantly triglycine; Irimia et al., 2014). These and Santy, 2012), their cognate mRNAs are also alternatively values may reflect a need for constitutive versus growth factor spliced, and the specific roles of the resulting isoforms have not dependent activation of Arf proteins in different tissues. To our been examined in the context of RTK signaling. Using multiple knowledge, our study is the first to characterize a functional approaches, including structure function, live cell imaging, and difference between cytohesin-1 splice variants. This points to a CRI​SPR/Cas9-mediated genetic deletion, we find that a splice key role for these proteins and their splicing in normal develop- variant of cytohesin-1 lacking a 3-nt microexon is specifically ment and disease. required for HGF activation of Met RTK-dependent cytoskeletal HGF stimulation of Met promotes PI3K activation and recruit- dynamics and cell migration. ment of diglycine cytohesin-1 to the leading edge for prolonged Alternative splicing diversifies the number of possible tran- periods (60 min). These data are consistent with an HGF-depen- scripts from a single . Indeed, it is believed that ∼95% of dent rapid and prolonged recruitment of PI3K to a Met signaling multiexon undergo alternative splicing (Pan et al., 2008). complex at the plasma membrane and generation of PI(3,4,5)P3 While significant effort has been put into establishing the regu- at the leading edge of migrating cells (Maroun et al., 1999a,b; latory mechanisms of alternative splicing, the functional signif- Frigault et al., 2008; Abella et al., 2010). The Gab1 scaffold is the icance of many of these events remains unknown. Microexons major determinant for recruitment of the p85 adapter have been reported in both plants and metazoan (Beachy et al., and PI3K activation following HGF stimulation of Met (Maroun 1985; Guo and Liu, 2015), and their role in generating alternative et al., 1999b). Notably, in contrast to HGF, EGF stimulation pro- transcripts is becoming increasingly appreciated (Ustianenko et motes a transient increase in PI3K activity associated with Gab1 al., 2017). Many human diseases are characterized in part by de- (Maroun et al., 1999b), and in response to EGF, cytohesin-1 is fects or alterations in alternative splicing patterns (Baralle and transiently recruited to the plasma membrane, supporting our Giudice, 2017), with emerging evidence suggesting the involve- model that diglycine cytohesin-1 is an important PI3K effector ment of microexon splicing in their etiology. For example, micro- (Venkateswarlu et al., 1999). While the role for PI3K in cell migra- exons have been implicated in autism spectrum disorders, which tion has been examined extensively, many studies have focused is in keeping with their abundance in neuron-specific transcripts on activation of the serine/threonine kinase Akt downstream

(Irimia et al., 2014). In a small cohort of symptomatic neuroblas- of PI(3,4,5)P3 in cell proliferation and survival (Fruman et al., toma patients, a highly phosphorylated neuronal microexon– 2017). It is now understood that the modulation of cell migra- containing splice variant of the non-RTK Src is lost (Brugge et tion by PI3K may predominantly involve Akt-independent path- al., 1985; Pyper and Bolen, 1990; Matsunaga et al., 1993; Keenan et ways that are not well understood (Lien et al., 2017). PI(3,4,5)P3 al., 2015). In general, however, information on the roles of micro- is recognized by multiple GEFs, including DOCK180, Vav2, and exon splicing in cancer is very limited. This work defines a func- P-REX1, that activate Rac to promote cell migration (Côté et al., tional role for microexon alternative splice variants in cancer cell 2005; Ménard et al., 2014; Graziano et al., 2017). Arf6 also acts

with DMSO, 0.2 µM Wortmannin, or 20 µM LY294002 for 30 min before HGF stimulation. (E) HeLa cells stably expressing triglcyine EGFP-CYTH1 were pretreated with 10 µM DMSO or ionomycin for 10 min, permeabilized with ice-cold 0.05% saponin in Pipes buffer, and imaged by confocal microscopy. Scale bars, 20 µm.

Ratcliffe et al. Journal of Cell Biology 293 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 upstream of the Rac1 GEF DOCK 180 to regulate cell migration from Cell Signaling. I(1,4,5)P3 (Q-0145), I(1,3,4,5)P4 (Q1345), and (Santy et al., 2005; Koubek and Santy, 2018). PI(3,4,5)P3 PolyPIPosomes (Y-P039) were obtained from Eche- Within this context, our findings establish a function for lon Biosciences. Wortmannin (W-2990) was obtained from LC the evolutionarily conserved alternatively spliced microexon in Laboratories, LY294002 (S1105) from Selleckchem, and ion- cytohesin-1 in phosphoinositide signaling, membrane dynamics omycin calcium salt from Streptomyces conglobatus (I0634) and cancer cell migration. The in vivo relevance of microexons from Sigma Aldrich. AllStars Neg. Control siRNA (1027281) is only beginning to be understood and may have wide was obtained from QIA​GEN. siRNA targeting Arf1 (5′-ACG​UGG​ ranging implications from normal development, neurological AAA​CCG​UGG​AGUA-3′), Arf3 (5′-UAU​GAA​CGC​UGC​UGA​GAUC- disease, and cancer. 3′), and Arf6 (5′-GAU​GAG​GGA​CGC​CAU​AAUC-3′) were obtained from Dharmacon. SMA​RTpool siRNA against human CYTH1 (9267: 5′-GGA​AUC​AUC​CCU​UUA​GAGA-3′, 5′-AAG​AGA​CGC​UGG​ Materials and methods UUC​AUUC-3′, 5′-GGG​AGA​GAG​AUG​AGU​UUAA-3′, 5′-GCA​AUA​ Cell culture and transfection AUG​GCG​UGU​UCCA-3′), CYTH2 (9266: 5′-GCA​AUG​GGC​AGG​ HeLa and 293T cell lines were cultured under standard AAG​AAGU-3′, 5′-AAA​CCG​AAC​UGC​UUU​GAAC-3′, 5′-UGG​CAG​ conditions at 37°C and 5% CO2 in 10% FBS. CYTH1 KO clonal lines UGC​UCC​AUG​CUUU-3′, 5′-GCC​AAU​GAG​GGC​AGU​AAGA-3′),

were generated using the lentiCRISPR​ v2 system (Sanjana et al., CYTH3 (9265: 5′-GGA​GAA​GGC​CUA​AAU​AAGA-3′, 5′-CAG​CAG​ Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 2014). Briefly, phosphorylated and annealed CYTH1-specific AGA​UCC​CUU​CUAU-3′, 5′-GAA​GAC​CUC​UCA​UUA​GAAG-3′, 5′- guide RNAs were cloned into the lentiCRI​SPRv2 vector using GGA​AUC​AUC​CCG​UUG​GAAA-3′), IQS​EC1 (9922: 5′-GGA​AGA​ BsmBI restriction sites. Lentiviral particles were produced by AAU​UCA​CCG​AUGA-3′, 5′-CGA​GAA​AUC​UUC​CUG​UUCA-3′, Lipo2000 transfection of 293T cells with EGFP-CYTH1, psPAX2, 5′-GGA​CGA​UGG​UGA​GGA​CAUU-3′, GAC​AGU​CCU​UCU​CCU​ and pMD2.G vectors according to the manufacturer’s protocol. UGUA-3′), IQS​EC2 (23096: 5′-GAA​AUA​CCG​UGU​GGA​GUCG-3′, Filtered supernatant was then used to infect HeLa cells, and 5′-GUA​CAG​UCC​CAG​CGU​CAAA, 5′-CCC​AGA​AAG​UGG​AGC​GACU- 24 h after infection, cells were selected in puromycin (P7255; 3′, 5′-UCU​GUA​GAG​UUG​UGG​CCAA-3′) and PSD3 (23362: 5′-CAA​ Sigma Aldrich) for 2 d. Clonal populations were established by CGA​AUU​UAG​CAA​ACUA, 5′-GGA​CGU​CGA​UGA​GUA​CAAA-3′, limiting dilution and screened for cytohesin-1 expression by 5′-GAC​CAU​CAG​UCG​UAU​UGGA-3′, 5′-UGU​CAG​GAG​UUC​AUU​ Western blot. To establish stable cell lines expressing EGFP- GCAA-3′) were obtained from Dharmacon. All the following CYTH1 diglycine and triglycine isoforms, as well as mutants, oligonucleotides were obtained from Integrated DNA Technol- triglycine EGFP-CYTH1 was first PCR amplified and subcloned ogies. DNA oligonucleotides used for quantitative RT-PCR were into a pLVX-IRES-Hyg vector. NEB Q5 site-directed mutagenesis CYTH1_FWD(5′-CAG​TGA​CCT​GAC​AGC​AGA​GG-3′) and REV(5′- was used to generate protospacer adjacent motif silent TCT​GAA​TGT​CAG​CCA​GCA​GC-3′); CYTH2_FWD(5′-AAC​CTG​TAC​ mutations, the diglycine isoform, as well as cytohesin-1 mutants. GAC​AGC​ATC​CG-3′′) and REV(5′-CCG​GTC​CGG​GTT​GAA​GAAG-3′); Lentiviral particles were produced by Lipo2000 transfection CYTH3_FWD(5′-CGT​GCC​TGA​AGA​CCT​CTC​AT-3′) and REV(5′- of EGFP-CYTH1, psPAX2, and pMD2.G vectors into 293T lines. TCG​TTT​TGC​TCT​CCT​CTA​CGG-3′); IQS​EC1_FWD(5′-GCA​CAG​GAT​ Filtered supernatant containing lentiviral particles was then AGAGTC​ GGA​ GC-3′)​ and REV(5′-CCCGAC​ TCC​ TTT​ TTG​ AGG​ CT-3′);​ concentrated by adding 1 vol PEG8000 to 3 vol supernatant, IQS​EC2_FWD(5′-TCC​AGT​CCC​ATA​TCC​GGG​TT-3′) and REV(5′- incubating overnight at 4°C, centrifuging at 2,750 g for 30 min GAG​GGC​TGG​GTT​ACA​GAC​AC-3′); IQS​EC3_FWD(5′-CTA​CCA​CTG​ at 4°C, and resuspending the pellet in DMEM. HeLa clonal CGAGAA​ CCC​ AG-3′)​ and REV(5′-GATGCC​ CTT​ GTC​ GGG​ GTT​ TA-3′);​ population 1 expressing lentiCRI​SPR v2 gRNA3 was infected, and PSD_FWD(5′-GGC​TGT​ACC​GAC​TAG​ATG​GC-3′) and REV(5′-AGC​ 2 d later, stable cell lines were selected and cultured in 10% FBS in TTG​GTC​CAG​AGT​CAT​GC-3′); PSD2_FWD(5′-ACC​CTG​ATG​ACA​ DMEM and 600 µg/ml Hygromycin (10687010; Thermo Fisher). GCA​CTT​CG-3′) and REV(5′-TTT​TGC​CAA​TGT​TGT​GGC​CG-3′); mCherry2-PTEN was generated by PCR amplifying mCherry2 PSD3_FWD(5′-AGC​GTG​GCA​CAT​GAA​CAA​AC-3′) and REV(5′-CCT​ (plasmid 54563; Addgene) and subcloning into pcDNA3 GFP- CGA​CCC​TTC​CCC​TAG​AA-3′); and PSD4_FWD(5′-TTG​GAG​GCC​ PTEN (plasmid 10759; Addgene) between BamHI and HindIII ATG​TTT​GGG​TC-3′) and REV(5′-CAC​ACT​GAC​ACA​CCT​CCC​TC- restriction sites. NEB Q5 site-directed mutagenesis was used 3′). DNA oligonucleotides used for determination of cytohesin to generate the C124S mutant. Stable cell lines expressing family microexon splicing patterns were CYTH1_TIDE_FWD(5′- mCherry2-PTEN WT or C124S were generated by transfecting GCA​GAG​GAG​CGT​CAA​GAA​CT-3′), CYTH1_TIDE_REV(5′-CGT​AGA​ pcDNA3 vectors and, after 48 h, culturing in standard media with AAG​GGT​CCC​TGC​TG-3′), CYTH1_TIDE_seq(5′-GAT​AAG​CCC​ACT​ 1.2 mg/ml G418. After 1 wk, mCherry2-positive cells were sorted GTG​GAG​AGG​TT-3′); CYTH2_TIDE_FWD(5′-GAG​GAC​GGC​GTC​ by FACS, and the mCherry2+ population was expanded. TAT​GAA​CC-3′), CYTH2_TIDE_REV(5′-TGC​TCC​TGC​TTC​TTC​TTG​ ACT-3′), CYTH2_TIDE_seq(5′-GAC​AAG​CCG​GGC​CTG​GAG​CGC​ Antibodies and reagents TT3′); CYTH3_TIDE_FWD(5′-CGT​GCC​TGA​AGA​CCT​CTC​AT-3′), Commercial antibodies to cytohesin-1 (2E11; mouse MA1-060) CYTH3_TIDE_REV(5′-TCC​TCG​TTG​CCA​ACA​TGT​CA-3′) and were obtained from Thermo Fisher; GAP​DH (rabbit sc-25778) CYTH3_TIDE_seq(5′-GAC​AAG​CCC​ACG​GCA​GAA​CGG​TT-3′). DNA from Santa Cruz; GFP (rabbit A6455) from Life Technologies; oligonucleotides used to generate CYTH1 gRNA1_FWD(5′-CAC​ and Akt (mouse 2920S), p-Akt1 (rabbit 9018S), p44/42 MAPK CGG​AAC​ATC​CGA​CGG​AGA​AAAC-3′) and REV(5′-AAA​CGT​TTT​ (mouse 9107S), p-p44/42 MAPK (T202/Y204; rabbit 9101L), Met CTC​CGT​CGG​ATG​TTCC-3′); and CYTH1 gRNA3_FWD(5′-CAC​CGG​ (mouse 3127L), p-Met (rabbit 3077S), and PTEN (rabbit 9188S) CAG​CTC​CTG​TTT​TCT​CCGT-3′) and _REV(5′-AAA​CAC​GGA​GAA​

Ratcliffe et al. Journal of Cell Biology 294 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 AAC​AGG​AGC​TGCC-3′). DNA oligonucleotides used to subclone view were then captured per condition, and 10 cells were tracked cyothesin-1 were CYTH1_Xho1_EGFP FWD(5′-AAA​ACT​CGA​GAT​ per field of view. Tracks between 16 and 24 h after stimulation GGT​GAG​CAA​GGG​CGA​GGA​GC-3′), CYTH1_Xba1 Rev(AAA​ATC​ were used for quantification. Cells were tracked using the track TAG​ATC​AGT​GTC​GCT​TCG​TGG​AGG); CYTH1_BamH1-PH_257_ points application in MetaMorph (Molecular Devices). FWD(5′-AAA​AGG​ATC​CAC​GTT​CTT​CAA​CCC​AGA​CCG​AG-3′) and CYTH1_EcoR1-PH_381_REV (5′-TTT​TGA​ATT​CCT​AGA​ Collagen invasion assays AAG​GGT​CCC​TGC​TGA​TGG​CTG​CTT​TAA​TGC-3′); and CYTH1_ Invasion assays were performed as previously described, with BamH1-PH_243 _FWD(5′-AAA​AGG​ATC​CCC​CTT​TAA​AAT​CCC​ minor modifications (Knight et al., 2018). Collagen gels (3 mg/ml AGA​AGAC-3′) and CYTH1_EcoR1-PH_397_REV(5′-TTT​TGA​ATT​ in PBS, pH 7.0–7.5) were polymerized in chambers of an eight- CTC​AGT​GTC​GCT​TCG​TGG​AGGA-3′). DNA oligonucleotides used well chamber slide. 5 × 104 cells were seeded, and these were to subclone mCherry2 were HindIII-mCherry2_FWD(5′-AAA​ overlayed with an additional layer of collagen. Media without or AAA​GCT​TGC​CAC​CAT​GGT​GAG​CAA​GGG​CGA​GGA​GG-3′) and without HGF was then added, and cells were grown for 7 d before BamHI-mCherry2_REV(5′-TTT​TGG​ATC​CCT​GAA​GCT​GCT​GCT​ fixation. Media was changed every 2 d. Fixed collagen gels were TGT​ACA​GCT​CGT​CCA​TGCC-3′). DNA oligonucleotides used for embedded in optimal cutting temperature compound and imme- site directed mutagenesis were CYTH1_gRNA3_PAM_SDM_FWD diately frozen in liquid nitrogen. 8-µm sections were mounted

(5′-TGGAGA​ ACA​ TTC​ GAC​ GGA​ GAAA-3′)​ and REV(5′-GTTCTT​ GAC​ ​ on poly-L-lysine–coated microscope slides and stained with DAPI Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 GCT​CCT​CTG​CTG-3′); CYTH1_3Gto2G_SDM_FWD(5′-TGG​CAG​ before mounting with a glass coverslip. Sections were imaged at GGT​AAA​GAC​TTG​GAA​GAG-3′) and REV(5′-CCG​AGT​TTC​AAT​AGC​ 10× magnification using the DAPI light cube on a EVOS FL Cell CAG​CCTT-3′); CYTH1_E157K_SDM_FWD(5′-GCT​ACC​CGG​AAA​ Imaging System. GGC​CCA​GAA-3′) and REV(5′- CGG​AAG​CTC​CAC​AGG​AAC​TG-3′); CYTH1_R281C_SDM_FWD(5′-TTG​GAA​GAG​ATG​CTG​GTT​CAT​ Kymograph analysis TC-3′), GGG_R281C_REV(5′-GTC​TTT​ACC​CTG​CCA​CCTC-3′) and 24 h after plating 7,500 cells per well in a collagen-coated 24- GG_R280C_REV(5′-GTC​TTT​ACC​CTG​CCA​CCG-3′); PTEN_C124S_ well dish, cells were rinsed twice with 0.02% FBS in DMEM, and FWD(5′-AGC​AAT​TCA​CGC​TAA​AGC​TGG​AAA​GGG-3′) and PTEN_ 0.9 ml 0.02% FBS in DMEM was added. Cells were replaced in the C124S_REV(5′-GCA​ACA​TGA​TTG​TCA​TCT​TC-3′). All vectors were incubator overnight. The following morning, 0.1 ml 0.02% FBS in sequenced verified before use. DMEM or 0.5 nM HGF in 0.02% FBS in DMEM was added to each well. To analyze membrane dynamics, the dish was transferred Microexon splicing patterns to and imaged with a Axiovert 200 M inverted microscope, LD Microexon PSI values were determined by adapting a protocol A-Plan 20×/0.3 Ph1 objective lens, and AxioCam HRM; all were designed for estimating the frequency of genome editing events contained within a transparent environment chamber Climabox

(Brinkman et al., 2014). cDNA from HeLa, MDA-MB-231, and maintained at 5% (vol/vol) CO2 at 37°C and driven by AxioVision HCC1143 cells was generated using the transcriptor first-strand LE software. The motorized stage was preprogrammed to ad- cDNA synthesis kit (Roche). PCR reactions using cDNA as a vance to defined locations, and images were captured every 15 s, template and primers flanking the microexon were performed beginning 15 min after HGF addition for 1 h. Images generated using Dreamtaq green PCR master mix (Thermo Fisher). PCR between 15 and 60 min after HGF stimulation were used to gener- conditions were 1 min at 95°C (once); 30 s at 95°C, 30 s at 55°C, ated kymographs using the kymograph function in MetaMorph. and 1 min at 72°C (35 times); and 5 min at 72°C (once). Amplicons were purified and sent for Sanger sequencing. The resulting Analysis of actin rearrangement chromatograms were analyzed using TIDE software (https://tide​ ​ For analysis of the actin cytoskeleton, cells were prepared as .deskgen​.com) to determine PSI values. for kymograph analysis, except 1 h after HGF addition, the cells were fixed with 4% PFA. Coverslips were then processed and Cell migration counterstained with Alexa Fluor 546–labeled Phalloidin and Wells of a 24-well dish were coated with 25 µg/ml collagen for DAPI. Images were acquired using a 63× Plan Apochromat, 1.40 1 h at 37°C and washed twice with PBS. Cells were plated at 7,500 NA oil immersion objective on a LSM 800 laser scanning confocal cells per well in collagen-coated wells and, where indicated, microscope (Carl Zeiss) driven by ZEN Blue software (Carl Zeiss). immediately transfected with 20 nM siRNA using HiPerfect 10 independent fields of view were chosen per condition, and per the manufacturer’s instructions. Assays were performed cells at the periphery of colonies were manually scored for actin 24–48 h after plating. Media was aspirated and replaced with rich ruffles at the periphery of the cell. More than 100 cells per growth media or growth media containing 0.5 nM HGF. The dish condition were scored. was then transferred to and images captured with an Axiovert 200 M inverted microscope (Carl Zeiss), LD A-Plan 20×/0.3 PH domain purification Ph1 objective lens, and AxioCam HRM (Carl Zeiss); all were All recombinant proteins were expressed using Rosetta-2(DE3) contained within a transparent environment chamber Climabox Escherichia coli cells (EMD Biosciences) and purified using

(Carl Zeiss) maintained at 5% (vol/vol) CO2 at 37°C and driven Glutathione Sepharose 4B resin (GE Healthcare). Bacterial by AxioVision LE software (Carl Zeiss). The motorized stage was cultures were grown at 37°C to an OD600 of 0.7–08 and then preprogrammed to advance to defined locations, and images induced using 1 mM IPTG and incubated overnight at 25°C. were captured every 15 min for 24 h. Three independent fields of The cells were harvested using a JLA-10.1 rotor at 6,000

Ratcliffe et al. Journal of Cell Biology 295 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 relative centrifugal force. The cell pellets were resuspended Imaging subcellular localization of EGFP-cytohesin-1 in buffer A (50 mM Na/K phosphate buffer, pH 8.0, 150 mM Assays were performed 48 h after plating 7.5 × 104 cells in an ibidi NaCl, 0.1% β-mercaptoethanol, and 1× protease inhibitor glass-bottom dish (81158). Media was aspirated, and cells were cocktail; Roche). 200 µg/ml Lysozyme and 10 U/ml DNase1 rinsed twice with 0.02% FBS in DMEM, and 1.35 ml 0.02% FBS in was added to the cells and, after a 30-min incubation on ice, DMEM was added. Cells were replaced in the incubator for 2–3 h. samples were lysed by sonication, followed by addition of 0.5% A bolus of 5 nM HGF was added to each plate (Cf = 0.5 nM). After Triton X-100. Lysates were then cleared by centrifugation at the indicated time points, media was aspirated and ice-cold 0.05% 4°C (JA-20; 35 min at 35,000 relative centrifugal force) and saponin in Pipes buffer (80 mM Pipes KOH, pH 7.0, 5 mM EGTA, filtered using a 0.45-μM filter. Glutathione Sepharose resin and 1 mM MgCl2) was added. Cells were imaged immediately after was added to the soluble fraction followed by incubation at the cytosolic fraction had dissipated (∼2–5 min). When cells were 4°C for 1 h and then three washes with buffer A. GST-tagged treated with DMSO or inhibitor, this was added 20 min before PH domains were then eluted with 50 mM reduced glutathione HGF stimulation (Wortmannin and LY294002) or 10 min before in buffer A, and 50 U Precission Protease (GE Healthcare) permeabilization (ionomycin). Images were acquired using a 63× was added to the supernatant and the sample was dialyzed Plan Apochromat, 1.40 NA oil immersion objective on a LSM 800 using a 3,500–molecular weight cut-off membrane (Spectrum laser scanning confocal microscope driven by ZEN Blue software.

Laboratories) against buffer B (25 mM Tris, 100 mM NaCl, 1 mM Using MetaMorph, a linescan with a width of 10 pixels was then Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 DTT, and 1 mM EDTA, pH 7.3) overnight at 4°C. The sample manually drawn along the perimeter of the cell, starting from was then incubated with Glutathione Sepharose resin for 1 h the innermost point of the cell relative to the colony. Linescan to remove cleaved GST, followed by separation of the resin length was normalized and divided into 24 equal sections. The and concentration of the sample using an Amicon 10K unit mean fluorescence of each section was averaged across cells and (Millipore). The concentrated sample was then applied to a plotted using the polar histogram function in MATLAB.​ Sephadex S75 column (GE Healthcare) using MOPS (100 mM NaCl, pH 7.5) as the running buffer. Sample purity was assessed SDS-PAGE and immunoblotting by SDS-PAGE, and the purest fractions were pooled for further HeLa cells were lysed in Triton X-100-glycerol-Hepes lysis experiments (Fig. S4 B). The concentration of the sample was buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl2, 1 mM calculated using the Bio-Rad Bradford Protein Assay. EGTA, 1% Triton X-100, 10% glycerol, 1 mM PMSF, 1 mM sodium vanadate, 10 µg/ml aprotinin, and 10 µg/ml leupeptin). Equal ITC amounts of protein were aliquoted, and SDS sample buffer was All ITC experiments were performed on a MicroCal added and boiled for 5 min. Samples were then resolved by SDS- microcalorimetry system (GE Healthcare). PH domains, PAGE and transferred to Immobilon-FL polyvinylidene fluoride phosphoinositide liposomes (Echelon Bioscience), and inositol transfer membranes. Membranes were blocked with blocking phosphates (Echelon Biosciences) were prepared in 1× MOPS. buffer (LI-COR Biosciences) and incubated with primary To measure binding kinetics to lipososomes, the reaction cell antibodies, followed by incubation with infrared conjugated contained 17.5 µM of PI(3,4,5)P3-containing liposomes and was secondary antibodies before detection on the Odyssey IR Imaging titrated with 155 µM of recombinant CYTH1. To measure binding System (LI-COR Biosciences). kinetics to inositol phosphates, the reaction cell was filled with 25 µM CYTH recombinant protein and the sample was titrated Quantitative RT-PCR with either I(1,3,4,5)P4 or I(1,4,5)P3 (Echelon Biosciences). All Total RNA was isolated using QIAGEN​ AllPrep DNA/RNA mini kit experiments were performed at 20°C, the reaction cell contained as per the manufacturer’s instructions. 1,000 ng of total RNA was 320 µl of sample and 70 µl of the titrant in the syringe, which used for QIA​GEN OneStep RT-PCR kit as per the manufacturer’s was set between 19 and 38 injections at 2.5–1.25 µl per injection. instructions, and data were collected and analyzed using a Roche The binding isotherm was fitted with a model that uses a single LightCycler 480. Data were normalized to GAPDH.​ set of independent sites to determine the stoichiometry and thermodynamic binding constant. Quantification and statistical analysis Quantitative data are presented as the means ± SEM. Statistical Modeling of the CYTH1 PH domain significance was assessed using a two-tailed Student’s t test or The 3D protein structural model of the diglycine CYTH1 PH do- two-way ANO​VA where indicated in the figure legend. P values main was generated using the methods described in Swiss-Model and the number of experiments used for quantification and sta- (https://​swissmodel​.expasy​.org/​; Arnold et al., 2006). After tar- tistical analysis are indicated in the corresponding figure legends. get and template selection using the software, the final CYTH1 model was built against the structure of the PH domain from cyto- Online supplemental material hesin-3 ( accession numbers 2R09 and 2R0D). Fig. S1 shows the impact of Arf1, Arf3, or Arf6 depletion on The ligand present in the template structure was transferred by HGF-dependent remodeling of the actin cytoskeleton. Fig. S2 homology to the model. Additional analysis of the binding pocket shows the expression of Arf GEFs in HeLa cells, CYTH family with the I(1,4,5)P3 and I(1,3,4,5)P4 ligands was performed by member expression in MDA-MB-231 and HCC1143 cells, and the overlaying the model structure to the PH domains from ARNO efficiency of siRNA-mediated silencing. It also shows the pro- (Protein Data Bank accession numbers 1U27 and 1U29). tein levels of cytohesin-1 in control and CYTH1 KO lines. Fig. S3

Ratcliffe et al. Journal of Cell Biology 296 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 shows Met protein stability and downstream signaling in re- Chardin, P., S. Paris, B. Antonny, S. Robineau, S. Béraud-Dufour, C.L. Jack- son, and M. Chabre. 1996. A human exchange factor for ARF contains sponse to HGF. Fig. S4 shows expression levels of overexpressed Sec7- and pleckstrin-homology domains. Nature. 384:481–484. https://​ EGFP-tagged cytohesin-1 isoforms and mutants as well as the pu- doi.org/​ 10​ .1038/​ 384481a0​ rified PH cytohesin-1 PH domains. Fig. S5 shows the localization Côté, J.-F., A.B. Motoyama, J.A. Bush, and K. Vuori. 2005. A novel and evo- of diglycine GFP-CYTH1 in PTEN-overexpressing cells. Video 1 lutionarily conserved PtdIns(3,4,5)P3-binding domain is necessary for DOCK180 signalling. Nat. Cell Biol. 7:797–807. https://​doi​.org/​10​.1038/​ shows the localization of diglycine EGFP-CYTH1 upon saponin ncb1280 permeabilization in HGF-treated cells. Cronin, T.C., J.P. DiNitto, M.P. Czech, and D.G. Lambright. 2004. Structural determinants of phosphoinositide selectivity in splice variants of Grp1 family PH domains. EMBO J. 23:3711–3720. https://​doi​.org/​10​.1038/​sj​ .emboj.7600388​ Acknowledgments Donaldson, J.G., and C.L. Jackson. 2011. ARF family G proteins and their regu- We thank members of the Park laboratory and Harvey W. Smith lators: roles in membrane transport, development and disease. Nat. Rev. Mol. Cell Biol. 12:362–375. for their helpful comments on the manuscript, Genentech for El azreq, M.-A., and S.G. Bourgoin. 2011. Cytohesin-1 regulates human blood HGF, Dr. Audrey Claing and Dr. Pierre-Luc Boulay for triglycine neutrophil adhesion to endothelial cells through β2 integrin activation. EGFP-CYTH1 cDNA, and Dongmei Zuo and the McGill Advanced Mol. Immunol. 48:1408–1416. Bioimaging Facility for technical assistance. Eyster, C.A., J.D. Higginson, R. Huebner, N. Porat-Shliom, R. Weigert, W.W. Wu, R.-F. Shen, and J.G. Donaldson. 2009. Discovery of new cargo pro-

This research was supported by the Fonds de Recherche du teins that enter cells through clathrin-independent endocytosis. Traffic. Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 Québec – Santé (doctoral studentships to P.P. Coelho and C.D.H. 10:590–599. Ratcliffe) and the Rosalind Goodman Commemorative Schol- Frigault, M.M., M.A. Naujokas, and M. Park. 2008. Gab2 requires membrane targeting and the Met binding motif to promote lamellipodia, cell scat- arship (C.D.H. Ratcliffe) and Canadian Institutes of Health Re- ter, and epithelial morphogenesis downstream from the Met receptor. search foundation operating grants 148423 (N. Sonenberg) and J. Cell. Physiol. 214:694–705. 242529 (M. Park). M. Park holds the Diane and Sal Guerrera Chair Frittoli, E., A. Palamidessi, P. Marighetti, S. Confalonieri, F. Bianchi, C. Mal- inverno, G. Mazzarol, G. Viale, I. Martin-Padura, M. Garré, et al. 2014. A in Cancer Genetics. RAB5/RAB4 recycling circuitry induces a proteolytic invasive program The authors declare no competing financial interests. and promotes tumor dissemination. J. Cell Biol. 206:307–328. Author contributions: Conceptualization, C.D.H. Ratcliffe and Fruman, D.A., H. Chiu, B.D. Hopkins, S. Bagrodia, L.C. Cantley, and R.T. Abra- M. Park; Methodology, C.D.H. Ratcliffe and N. Siiddiqui; Formal ham. 2017. The PI3K Pathway in Human Disease. Cell. 170:605–635. https://doi​ .org/​ 10​ .1016/​ j​ .cell​ .2017​ .07​ .029​ Analysis, C.R., N. Siiddiqui, P.P Coelho, and T.N. Cookey; Investi- Fukaya, M., S. Ohta, Y. Hara, H. Tamaki, and H. Sakagami. 2016. Distinct sub- gation, C.D.H. Ratcliffe, N. Siddiqui, P.P Coelho, and N. Laterreur; cellular localization of alternative splicing variants of EFA6D, a guanine Resources, N. Sonenberg and M. Park; Writing – Original Draft, nucleotide exchange factor for Arf6, in the mouse brain. J. Comp. Neurol. 524:2531–2552. https://doi​ .org/​ 10​ .1002/​ cne​ .24048​ C.D.H. Ratcliffe; Writing – Review & Editing, C.D.H. Ratcliffe, N. Gherardi, E., W. Birchmeier, C. Birchmeier, and G. Vande Woude. 2012. Target- Siddiqui, and M.P; Visualization, C.D.H. Ratcliffe and N. Siddiqui; ing MET in cancer: rationale and progress. Nat. Rev. Cancer. 12:89–103. Supervision, N. Sonenberg and M. Park. https://doi​ .org/​ 10​ .1038/​ nrc3205​ Gillingham, A.K., and S. Munro. 2007. The small G proteins of the Arf family and their regulators. Annu. Rev. Cell Dev. Biol. 23:579–611. https://doi​ .org/​ ​ Submitted: 17 April 2018 10.1146/​ annurev​ .cellbio​ .23​ .090506​ .123209​ Revised: 19 September 2018 Graziano, B.R., D. Gong, K.E. Anderson, A. Pipathsouk, A.R. Goldberg, and Accepted: 17 October 2018 O.D. Weiner. 2017. A module for Rac temporal signal integration revealed with optogenetics. J. Cell Biol. 216:2515–2531. https://doi​ .org/​ 10​ .1083/​ jcb​ ​ .201604113 Guo, L., and C.-M. Liu. 2015. A single-nucleotide exon found in Arabidopsis. References Sci. Rep. 5:18087. https://doi​ .org/​ 10​ .1038/​ srep18087​ Hashimoto, S., Y. Onodera, A. Hashimoto, M. Tanaka, M. Hamaguchi, A. Ya- Abella, J.V., C.A. Parachoniak, V. Sangwan, and M. Park. 2010. Dorsal ruffle mada, and H. Sabe. 2004. Requirement for Arf6 in breast cancer inva- microdomains potentiate Met receptor tyrosine kinase signaling and sive activities. Proc. Natl. Acad. Sci. USA. 101:6647–6652. https://doi​ .org/​ ​ down-regulation. J. Biol. Chem. 285:24956–24967. https://​doi​.org/​10​ 10.1073/​ pnas​ .0401753101​ .1074/jbc​ .M110​ .127985​ Arnold, K., L. Bordoli, J. Kopp, and T. Schwede. 2006. The SWI​SS-MOD​EL Hongu, T., Y. Funakoshi, S. Fukuhara, T. Suzuki, S. Sakimoto, N. Takakura, M. workspace: a web-based environment for protein structure homol- Ema, S. Takahashi, S. Itoh, M. Kato, et al. 2015. Arf6 regulates tumour ogy modelling. Bioinformatics. 22:195–201. https://​doi​.org/​10​.1093/​ angiogenesis and growth through HGF-induced endothelial β1 integrin bioinformatics/bti770​ recycling. Nat. Commun. 6:7925. https://doi​ .org/​ 10​ .1038/​ ncomms8925​ Baralle, F.E., and J. Giudice. 2017. Alternative splicing as a regulator of devel- Humphreys, D., V. Singh, and V. Koronakis. 2016. Inhibition of WAVE Regula- opment and tissue identity. Nat. Rev. Mol. Cell Biol. 18:437–451. tory Complex Activation by a Bacterial Virulence Effector Counteracts Beachy, P.A., S.L. Helfand, and D.S. Hogness. 1985. Segmental distribution Pathogen Phagocytosis. Cell Reports. 17:697–707. https://doi​ .org/​ 10​ .1016/​ ​ of bithorax complex proteins during Drosophila development. Nature. j.celrep​ .2016​ .09​ .039​ 313:545–551. https://doi​ .org/​ 10​ .1038/​ 313545a0​ Irimia, M., R.J. Weatheritt, J.D. Ellis, N.N. Parikshak, T. Gonatopoulos-Pour- Botelho, R.J., M. Teruel, R. Dierckman, R. Anderson, A. Wells, J.D. York, T. natzis, M. Babor, M. Quesnel-Vallières, J. Tapial, B. Raj, D. O’Hanlon, et Meyer, and S. Grinstein. 2000. Localized biphasic changes in phos- al. 2014. A highly conserved program of neuronal microexons is mis- phatidylinositol-4,5-bisphosphate at sites of phagocytosis. J. Cell Biol. regulated in autistic brains. Cell. 159:1511–1523. https://doi​ .org/​ 10​ .1016/​ ​ 151:1353–1368. https://doi​ .org/​ 10​ .1083/​ jcb​ .151​ .7​ .1353​ j.cell​ .2014​ .11​ .035​ Brinkman, E.K., T. Chen, M. Amendola, and B. van Steensel. 2014. Easy quan- Keenan, S., P.A. Lewis, S.J. Wetherill, C.J. Dunning, and G.J. Evans. 2015. The titative assessment of genome editing by sequence trace decomposition. N2-Src neuronal splice variant of C-Src has altered SH3 domain ligand Nucleic Acids Res. 42:e168. https://doi​ .org/​ 10​ .1093/​ nar/​ gku936​ specificity and a higher constitutive activity than N1-Src. FEBS Lett. Brugge, J.S., P.C. Cotton, A.E. Queral, J.N. Barrett, D. Nonner, and R.W. Keane. 589:1995–2000. https://doi​ .org/​ 10​ .1016/​ j​ .febslet​ .2015​ .05​ .033​ 1985. Neurones express high levels of a structurally modified, activated Klarlund, J.K., W. Tsiaras, J.J. Holik, A. Chawla, and M.P. Czech. 2000. Distinct form of pp60c-src. Nature. 316:554–557. https://doi​ .org/​ 10​ .1038/​ 316554a0​ polyphosphoinositide binding selectivities for pleckstrin homology do- Casanova, J.E. 2007. Regulation of Arf activation: the Sec7 family of guanine mains of GRP1-like proteins based on diglycine versus triglycine motifs. nucleotide exchange factors. Traffic. 8:1476–1485. J. Biol. Chem. 275:32816–32821. https://doi​ .org/​ 10​ .1074/​ jbc​ .M002435200​

Ratcliffe et al. Journal of Cell Biology 297 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106 Knight, J.F., R. Lesurf, H. Zhao, D. Pinnaduwage, R.R. Davis, S.M.I. Saleh, D. Oh, S.J., and L.C. Santy. 2012. Phosphoinositide specificity determines which Zuo, M.A. Naujokas, N. Chughtai, J.I. Herschkowitz, et al. 2013. Met syn- cytohesins regulate β1 integrin recycling. J. Cell Sci. 125:3195–3201. ergizes with p53 loss to induce mammary tumors that possess features https://doi​ .org/​ 10​ .1242/​ jcs​ .101683​ of claudin-low breast cancer. Proc. Natl. Acad. Sci. USA. 110:E1301–E1310. Palamidessi, A., E. Frittoli, M. Garré, M. Faretta, M. Mione, I. Testa, A. Diaspro, https://doi​ .org/​ 10​ .1073/​ pnas​ .1210353110​ L. Lanzetti, G. Scita, and P.P. Di Fiore. 2008. Endocytic trafficking of Rac Knight, J.F., V.Y.C. Sung, E. Kuzmin, A.L. Couzens, D.A. de Verteuil, C.D.H. is required for the spatial restriction of signaling in cell migration. Cell. Ratcliffe, P.P. Coelho, R.M. Johnson, P. Samavarchi-Tehrani, T. Gruosso, 134:135–147. https://doi​ .org/​ 10​ .1016/​ j​ .cell​ .2008​ .05​ .034​ et al. 2018. KIB​RA (WWC1) Is a Metastasis Suppressor Gene Affected Pan, Q., O. Shai, L.J. Lee, B.J. Frey, and B.J. Blencowe. 2008. Deep surveying by 5q Loss in Triple-Negative Breast Cancer. Cell Reports. of alternative splicing complexity in the human transcriptome by 22:3191–3205. https://doi​ .org/​ 10​ .1016/​ j​ .celrep​ .2018​ .02​ .095​ high-throughput sequencing. Nat. Genet. 40:1413–1415. https://doi​ .org/​ ​ Koubek, E.J., and L.C. Santy. 2018. ARF1 and ARF6 regulate recycling of GRA​ 10.1038/​ ng​ .259​ SP/Tamalin and the Rac1-GEF Dock180 during HGF-induced Rac1 acti- Parachoniak, C.A., and M. Park. 2012. Dynamics of receptor trafficking in tu- vation. Small GTPases. 9:242–259. morigenicity. Trends Cell Biol. 22:231–240. https://​doi​.org/​10​.1016/​j​.tcb​ Li, R., J.F. Knight, M. Park, and A.M. Pendergast. 2015. Abl Kinases Regulate .2012.02​ .002​ HGF/Met Signaling Required for Epithelial Cell Scattering, Tubulogen- Parachoniak, C.A., Y. Luo, J.V. Abella, J.H. Keen, and M. Park. 2011. GGA3 func- esis and Motility. PLoS One. 10:e0124960. https://doi​ .org/​ 10​ .1371/​ journal​ ​ tions as a switch to promote Met receptor recycling, essential for sus- .pone.0124960​ tained ERK and cell migration. Dev. Cell. 20:751–763. https://doi​ .org/​ 10​ ​ .1016/j​ .devcel​ .2011​ .05​ .007​ Lien, E.C., C.C. Dibble, and A. Toker. 2017. PI3K signaling in cancer: beyond Powelka, A.M., J. Sun, J. Li, M. Gao, L.M. Shaw, A. Sonnenberg, and V.W. Hsu. AKT. Curr. Opin. Cell Biol. 45:62–71. https://doi​ .org/​ 10​ .1016/​ j​ .ceb​ .2017​ .02​ ​ 2004. Stimulation-dependent recycling of integrin β1 regulated by ARF6 .007 and Rab11. Traffic. 5:20–36. https://​doi​.org/​10​.1111/​j​.1600​-0854​.2004​ Downloaded from http://rupress.org/jcb/article-pdf/218/1/285/865215/jcb_201804106.pdf by guest on 11 August 2020 Malek, M., A. Kielkowska, T. Chessa, K.E. Anderson, D. Barneda, P. Pir, H. .00150.x​ Nakanishi, S. Eguchi, A. Koizumi, J. Sasaki, et al. 2017. PTEN Regulates Pyper, J.M., and J.B. Bolen. 1990. Identification of a novel neuronal C-SRC exon PI(3,4)P2 Signaling Downstream of Class I PI3K. Mol. Cell. 68:566–580. expressed in human brain. Mol. Cell. Biol. 10:2035–2040. e10. Rajadurai, C.V., S. Havrylov, K. Zaoui, R. Vaillancourt, M. Stuible, M. Naujokas, Maroun, C.R., M. Holgado-Madruga, I. Royal, M.A. Naujokas, T.M. Fournier, D. Zuo, M.L. Tremblay, and M. Park. 2012. Met receptor tyrosine kinase A.J. Wong, and M. Park. 1999b. The Gab1 PH domain is required for lo- signals through a cortactin-Gab1 scaffold complex, to mediate invado- calization of Gab1 at sites of cell-cell contact and epithelial morphogen- podia. J. Cell Sci. 125:2940–2953. esis downstream from the met receptor tyrosine kinase. Mol. Cell. Biol. Ratcliffe, C.D.H., P. Sahgal, C.A. Parachoniak, J. Ivaska, and M. Park. 2016. Reg- 19:1784–1799. https://doi​ .org/​ 10​ .1128/​ MCB​ .19​ .3​ .1784​ ulation of Cell Migration and β1 Integrin Trafficking by the Endosomal Maroun, C.R., D.K. Moscatello, M.A. Naujokas, M. Holgado-Madruga, A.J. Adaptor GGA3. Traffic. 17:670–688. Wong, and M. Park. 1999a. A conserved inositol phospholipid binding Reviriego-Mendoza, M.M., and L.C. Santy. 2015. The cytohesin guanosine site within the pleckstrin homology domain of the Gab1 docking protein exchange factors (GEFs) are required to promote HGF-mediated renal is required for epithelial morphogenesis. J. Biol. Chem. 274:31719–31726. recovery after acute kidney injury (AKI) in mice. Physiol. Rep. 3:e12442. Matsunaga, T., H. Shirasawa, M. Tanabe, N. Ohnuma, H. Takahashi, and B. Royal, I., N. Lamarche-Vane, L. Lamorte, K. Kaibuchi, and M. Park. 2000. Ac- Simizu. 1993. Expression of alternatively spliced src messenger RNAs tivation of cdc42, rac, PAK, and rho-kinase in response to hepatocyte related to neuronal differentiation in human neuroblastomas. Cancer growth factor differentially regulates epithelial cell colony spreading Res. 53:3179–3185. and dissociation. Mol. Biol. Cell. 11:1709–1725. https://​doi​.org/​10​.1091/​ Mayor, S., R.G. Parton, and J.G. Donaldson. 2014. Clathrin-independent path- mbc.11​ .5​ .1709​ ways of endocytosis. Cold Spring Harb. Perspect. Biol. 6:a016758. https://​ Sanjana, N.E., O. Shalem, and F. Zhang. 2014. Improved vectors and ge- doi.org/​ 10​ .1101/​ cshperspect​ .a016758​ nome-wide libraries for CRI​SPR screening. Nat. Methods. 11:783–784. Ménard, L., P.J. Parker, and S. Kermorgant. 2014. Receptor tyrosine kinase https://doi​ .org/​ 10​ .1038/​ nmeth​ .3047​ c-Met controls the cytoskeleton from different endosomes via different Santy, L.C., K.S. Ravichandran, and J.E. Casanova. 2005. The DOCK180/Elmo pathways. Nat. Commun. 5:3907. https://doi​ .org/​ 10​ .1038/​ ncomms4907​ complex couples ARNO-mediated Arf6 activation to the downstream Moravec, R., K.K. Conger, R. D’Souza, A.B. Allison, and J.E. Casanova. 2012. activation of Rac1. Curr. Biol. 15:1749–1754. https://doi​ .org/​ 10​ .1016/​ j​ .cub​ ​ BRAG2/GEP100/IQSec1 interacts with clathrin and regulates α5β1 inte- .2005.08​ .052​ grin endocytosis through activation of ADP ribosylation factor 5 (Arf5). Simanshu, D.K., D.V. Nissley, and F. McCormick. 2017. RAS Proteins and Their J. Biol. Chem. 287:31138–31147. https://doi​ .org/​ 10​ .1074/​ jbc​ .M112​ .383117​ Regulators in Human Disease. Cell. 170:17–33. https://doi​ .org/​ 10​ .1016/​ j​ ​ .cell.2017​ .06​ .009​ Morishige, M., S. Hashimoto, E. Ogawa, Y. Toda, H. Kotani, M. Hirose, S. Wei, Stephens, L.R., K.T. Hughes, and R.F. Irvine. 1991. Pathway of phosphatidyli- A. Hashimoto, A. Yamada, H. Yano, et al. 2008. GEP100 links epidermal nositol(3,4,5)-trisphosphate synthesis in activated neutrophils. Nature. growth factor receptor signalling to Arf6 activation to induce breast can- 351:33–39. https://doi​ .org/​ 10​ .1038/​ 351033a0​ cer invasion. Nat. Cell Biol. 10:85–92. https://doi​ .org/​ 10​ .1038/​ ncb1672​ Tushir, J.S., and C. D’Souza-Schorey. 2007. ARF6-dependent activation of ERK Müller, R., C. Herr, S.K. Sukumaran, N.N. Omosigho, M. Plomann, T.Y. Riyahi, and Rac1 modulates epithelial tubule development. EMBO J. 26:1806– M. Stumpf, K. Swaminathan, M. Tsangarides, K. Yiannakou, et al. 2013. 1819. https://doi​ .org/​ 10​ .1038/​ sj​ .emboj​ .7601644​ The cytohesin paralog Sec7 of Dictyostelium discoideum is required for Ustianenko, D., S.M. Weyn-Vanhentenryck, and C. Zhang. 2017. Microexons: phagocytosis and cell motility. Cell Commun. Signal. 11:54. discovery, regulation, and function. Wiley Interdiscip. Rev. RNA. 8. Muralidharan-Chari, V., H. Hoover, J. Clancy, J. Schweitzer, M.A. Suckow, V. Venkateswarlu, K., F. Gunn-Moore, J.M. Tavaré, and P.J. Cullen. 1999. EGF-and Schroeder, F.J. Castellino, J.S. Schorey, and C. D’Souza-Schorey. 2009. NGF-stimulated translocation of cytohesin-1 to the plasma membrane of ADP-ribosylation factor 6 regulates tumorigenic and invasive prop- PC12 cells requires PI 3-kinase activation and a functional cytohesin-1 erties in vivo. Cancer Res. 69:2201–2209. https://​doi​.org/​10​.1158/​0008​ PH domain. J. Cell Sci. 112:1957–1965. -5472.CAN​ -08​ -1301​ Whitman, M., C.P. Downes, M. Keeler, T. Keller, and L. Cantley. 1988. Type I Ogasawara, M., S.-C. Kim, R. Adamik, A. Togawa, V.J. Ferrans, K. Takeda, M. phosphatidylinositol kinase makes a novel inositol phospholipid, phos- Kirby, J. Moss, and M. Vaughan. 2000. Similarities in function and gene phatidylinositol-3-phosphate. Nature. 332:644–646. https://​doi​.org/​10​ structure of cytohesin-4 and cytohesin-1, guanine nucleotide-exchange .1038/332644a0​ proteins for ADP-ribosylation factors. J. Biol. Chem. 275:3221–3230. Yoo, J.H., D.S. Shi, A.H. Grossmann, L.K. Sorensen, Z. Tong, T.M. Mleynek, https://doi​ .org/​ 10​ .1074/​ jbc​ .275​ .5​ .3221​ A. Rogers, W. Zhu, J.R. Richards, J.M. Winter, et al. 2016. ARF6 is an Oh, S.J., and L.C. Santy. 2010. Differential effects of cytohesins 2 and 3 on β1 in- actionable node that orchestrates oncogenic GNAQ signaling in uveal tegrin recycling. J. Biol. Chem. 285:14610–14616. https://doi​ .org/​ 10​ .1074/​ ​ melanoma. Cancer Cell. 29:889–904. https://doi​ .org/​ 10​ .1016/​ j​ .ccell​ .2016​ ​ jbc.M109​ .043935​ .04.015​

Ratcliffe et al. Journal of Cell Biology 298 Cytohesin-1 regulates HGF-dependent cell migration https://doi.org/10.1083/jcb.201804106